How to make a DFS root server the primary target?

[FreshPrince]

I have a problem with my DFS implementation.

I have 2 windows 2003 standard member servers with DFS. Let's call them DFS1 and DFS2

How do I force my users to access data on DFS1 via \\domain\dfs_share\ all the time, until it goes down, at which point, they should access DFS2 via \\domain\dfs_share\ ?

Right now, some users would be accessing DFS1 while others access DFS2 both via the same DFS share. As you can see, this is very dangerous and causes major file integrity issues. Someone opens an excel file and makes some changes, but finds out that those changes were over written by another person's modifications!

Is there any way to force everyone to go to the first server all the time, until that server goes down? I don't want anyone to access dfs via the 2nd server until there's a hardware failure.

Am I missing something here or is there a major flaw in the design of DFS?

Thanks for your help,

-FP

edit: little more info:

both DFS servers are root targets using the standard mesh replication topology. Everything else is default settings. I'm thinking about customizing the replication to one way from DFS1 to DFS2, but how do I still ensure that everyone hits DFS1 first?

 

[stash]

It's not a major flaw, it's called Conflict Resolution. For files in DFS, it is simple: last writer wins (folders are a little more complex).

FRS v1 (which does the actualy replication) is not designed for highly dynamic data, so if you have situations where people editing the same file at the same time on different shares is not acceptable, you may want to reconsider replicated DFS.

 

[FreshPrince]

Originally posted by: STaSh
It's not a major flaw, it's called Conflict Resolution. For files in DFS, it is simple: last writer wins (folders are a little more complex).

FRS v1 (which does the actualy replication) is not designed for highly dynamic data, so if you have situations where people editing the same file at the same time on different shares is not acceptable, you may want to reconsider replicated DFS.

Stash, how do you setup replicated DFS?

are you saying to change the replication topology or something else?

 

[stash]

Sounds like you are already using replicated DFS, based on your OP. Changing the replication topology won't help you with highly dynamic data.

 

[FreshPrince]

Originally posted by: STaSh
Sounds like you are already using replicated DFS, based on your OP. Changing the replication topology won't help you with highly dynamic data.

what do you recommend then? I have double take at another office, but it's too expensive for this office

 

[stash]

What is it that you are trying to accomplish? What is the point of having replicated DFS if you are just going to point everyone to one server (which afaik, you can't do)? If the server fails, what would you do then?

You are probably looking for some sort of redundancy, but DFS/FRS isn't going to be the answer with this type of data, imo. I don't know if any other product would be any better, though.

I'm not trying to be combative I'm just trying to get a sense of what you want to do and why you decided to use DFS.

 

[FreshPrince]

I'm just trying to have a simple file server for around 50-75 people that has some sort of redundancy. DFS, originally, I thought was great because the 2nd server would take over if the 1st one failed. However, due to data integrity problems with DFS, I can no longer use it. I guess double take it is then

 

[stash]

You may also want to consider using MS cluster for your file servers.

 

[azev]

I second clustering for your solution. But I think you may have to have windows 2003 enterprise for that, I am not so sure.

 

[FreshPrince]

yup, enterprise is required for clustering

grrr